jwdoherty Tue, 08/21/2007 - 17:34

Often refers to high speed routing that done using tags (or labels).

Routing is traditionally slower than switching because of the difficulty of finding the destination IP address quickly in a route table, where you often match against the best fitting address prefix, compared to finding a MAC address in a switch, usually an 1:1 exact match. (Newer technology, both hardware and software, narrows the speed differences.)

Consider though, if once you've done the intensive route match, you place a tag on the packet that indicates its forwarding path, then subsequent routers can lookup the path to forward the packet much, much faster; somewhat like a switch looks up a MAC.

MPLS is tag switching, although it refers to the same concept as label switching.

You might want to look at: http://www.cisco.com/univercd/cc/td/doc/cisintwk/ito_doc/tagswtch.htm

jwdoherty Wed, 08/22/2007 - 05:09

Other posters have discounted the performance advantages of the concept of tag/label switching, but as I originally noted although they don't have the impact they originally did, they are still somewhat present to some degree. (If I recall correctly, the primary motivation for the concept was also for performance.)

From http://www.cisco.com/univercd/cc/td/doc/cisintwk/ito_doc/mpls_tsw.htm

"From the previous description of the forwarding component, we can make several observations. First, the forwarding decision is based on the exact-match algorithm using a fixed-length, fairly short label as an index. This enables a simplified forwarding procedure, relative to longest-match forwarding traditionally used at the network layer.

This, in turn, enables higher forwarding performance (higher packets per second). The forwarding procedure is simple enough to allow a straightforward hardware implementation. A second observation is that the forwarding decision is independent of the label's forwarding granularity. The same forwarding algorithm, for example, applies to both unicast and multicast: A unicast entry would have a single (outgoing label, outgoing interface, outgoing link-level information) subentry, while a multicast entry might have one or more subentries. This illustrates how the same forwarding paradigm can be used in label switching to support different routing functions.

The simple forwarding procedure is thus essentially decoupled from the control component of label switching. New routing (control) functions can readily be deployed without disturbing the forwarding paradigm. This means that it is not necessary to reoptimize forwarding performance (by modifying either hardware or software) as new routing functionality is added."

Less work to accomplish a function is usually faster whether done in software or hardware, although with hardware the deltas often narrow. Also, often simpler function hardware is less expensive.

I will agee with the other posters that other features provided by MPLS have become more important.

mohammedmahmoud Tue, 08/21/2007 - 22:39


MPLS has a heritage stemming from Cisco's tag-switching protocol (Based upon Cisco's proprietary tag-switching protocol, the IETF has defined MPLS as a vendor-independent protocol), please do enjoy this document:


In brief they are both a way of switching packets based on labels/tags (fast way of packet switching on backbones, plus the extra available features that it introduces), review this document and it'll supply you with the elementary facts:



Mohammed Mahmoud.

royalblues Wed, 08/22/2007 - 00:20

Well the speed may have been a factor few years back but the new range of hardware devices are fast enough to do the route lookup in hardware and hence label or tag switching cannot be considered as a fast way to route traffic

The main difference would be to route packets based on labels rather than destination ip address (as in case of traditional routing). This way the core and the intermediate hops are not required to maintain the entire routing table end to end.



mohammedmahmoud Wed, 08/22/2007 - 00:49

Hi Narayan,

You are totally right, with the new hardware and the use of ASICs, speed is no more a strong benefit of running MPLS, only to add the MPLS added on features is one of the strongest benefit.

Chester, i'd strongly recommend that you review this book, its one of the very nice MPLS books that tackles MPLS from many prospectives.

MPLS Fundamentals

Luc De Ghein, CCIE

No. 1897


Mohammed Mahmoud.


This Discussion